在衍射分布的离散计算中，当衍射距离增大时，满足抽样定理的抽样点数会随之增加，为了解决补零算法带来的计算量大、计算机内存溢出等问题，提出了一种频移插值法，在不改变抽样点数的情况下，获得更为丰富的频谱成分，计算每次插值后的光场分布，然后进行线性叠加，从而得到完整的衍射分布。结果表明，该算法可以避开衍射距离增大必须增加抽样点数这一离散计算限制条件，和传统补零算法相比，计算量极大地减小，并降低了对计算机内存的要求。

In the discrete numerical calculation of diffraction distribution, the total number of sampling points following sampling theorem increases with the augment of propagating distance. Although the traditional zero-padding method resolves this problem, the calculation load increases inevitably and PC’s memory can not afford. A novel frequency shift interpolation, i.e., utilizing frequency spectrum calculation after shift in Fourier domain, was proposed. In the method, more sufficient spectrum components can be got without the increase of sampling number. A complete diffraction distribution is accomplished by splicing the light field obtained by spectrum of each shifted interpolation. The results show that the proposed method successfully evades the great sampling number caused by larger diffracting distance. Comparing with traditional zero-padding method, calculation load is decreased notably and the requirement to memory in numerical calculation is depressed.